On The Brink with Castle Island - Alex Svanevik (Nansen) on extracting signal from on-chain noise (EP.240)

Episode Date: September 6, 2021

We welcome Alex Svanevik, CEO of Nansen.ai to On the Brink to provide color on Nansen's unique approach to on-chain data and analysis. We discuss: The evolution of Nansen from a wallet labelling tool to an end-to-end analytics platform How Nansen provides actionable insights from its analysis of individual addresses, transactions, and contracts Nansen's process to scale the collection of individual wallet level data over time Specific ways end-users are leveraging Nansen to make data-driven NFT investments And more broadly, the application of AI to the analysis of blockchain data You can learn more about the company and its offerings at Nansen.ai and the company's YouTube channel.

Welcome to On the Brink with Castle Island. On today's episode, we had the honor to speak to Alex Svanavik, the CEO of on-chain analytics firm, nansen.a.I. Nansen has labeled more than 90 million addresses and counting on Ethereum and recently has started adding support for newer crypto networks. The firm superimposes the granular address and contract level information on more macro on-chain data to provide actionable insights on defy, NFTs, and other burgeoning categories in the crypto industry to its end users. In our conversation, Alex and I discuss the evolution of Nansen from a wallet labeling tool to an end-to-end analytics platform, how Nansen offers actionable insights from its analysis

of individual addresses and transactions, how the team has scaled the process of wallet labeling over time, specific ways that end users have been leveraging Nansen to make data-driven NFT investments, and more broadly, how Alex thinks that the application of AI will be complementary to the analysis of blockchain data longer term. Alex brings 10 years of experience an AI and data science prior to joining the crypto industry.

He's also an avid investor in Defi and NFTs himself, and in our conversation, shares a great deal of specific information about how he uses the data on Nansen. I learned a lot from our conversation and hope you walk away with some alpha yourself. So with that, let's turn to the discussion with Alex. Brought down by Bad Mortgage Investments, Lehman, which has 25,000 employees. will be liquidated. The federal government loans American International Group, AIG, $85 billion. This is a different kind of market, and the Fed is asleep. The federal government is stepping it to stabilize Fannie Mae and Freddie Mac, the two mortgage giants that have been threatened by the housing

crisis. The Bank of England has pumped 75 billion pounds more to Britain's ailing economy with a new round of quantitative easing. You print a couple trillion dollars and all of a sudden people start to worry. So out of this worry, we have something called the Bitcoin. Bitcoin. Hello, everyone. This is Ria from Castle Island Adventures, and I'm joined today by Alex Svanavik from nansen.a.I. Alex, thank you so much for joining us. It's great to have you on the podcast. Thanks for having me, Ria. Nice to be here. Awesome. Maybe just to kick off the discussion, could you tell us a little bit about yourself, your background pre-c crypto, how you got into crypto, and what you're building at Nansen now?

I've been working in the crypto space since 2017. I was initially drawn towards crypto through Ethereum. Like many people I had heard about Bitcoin many years before that, but I didn't really see the appeal, to be perfectly honest, with Bitcoin. But Ethereum was the thing that got me interested in the blockchain space. I was at the time working as a data science manager at a European media group. and after discovering Ethereum, it took me a couple of months before I resigned and started working full-time in crypto. I moved to Hong Kong, joined a startup there to build out a data team. And then right now I'm, of course, working on Nansen. So before that, I have a background in artificial intelligence and started a small consulting shop

back in 2010 trying to bring AI into the world of business, probably a bit too soon. And then I worked in management consulting for three years, everything from seafood to banking and insurance and luxury retail. And yeah, and then I moved into this European media group before I started working on crypto in 2017. And tell us a little bit about Nansen how you got the idea and maybe a little bit about the evolution as well because correct me if I'm wrong, but it started off. the idea was to build kind of like a wallet labeling API, but then as you learn more and fleshed out the idea more, you decided to convert it into an end-to-end data analytics platform. So tell us a little bit about that evolution. Yeah, that's exactly right.

So this was late 2019. I had already recruited Evgeny, who is now one of my co-founders at Nansen, to the startup I was working at in Hong Kong. So I already knew him, and he was working on an open source project called Ethereum ETL. And basically that was, I think, the first project that made it easy to query Ethereum data. So the idea is if you're an analyst, if you're a data scientist or engineer who wants to get Ethereum data into a more workable format, you could use Ethereum ETL to pull out the data from an Ethereum node and then start running SQL queries on it and so on. that project had actually been quite successful and again he worked with google to publish

Ethereum data as public data sets in BigQuery and so we were always there's maybe two angles on like how we ended up creating nonsense one angle was certainly that part where it was can we come up with a commercial business model on top of that project that's something that we had talked about for a long time and then from the other angle was like the demand that we saw in the market for better understanding of wallets and what people are actually doing on the blockchain and who are the different entities and so on and so forth. I had always been quite intrigued by the blockchain analytics companies like chain analysis, which was operating, it was kind of like a mysterious company, but I felt this is probably the data that they have is probably

useful for other purposes too, like investing, trading, market making and so on. And we had spoken to many people at different funds and so on who wanted better data on on chain activity so we figured let's build out the best database in the world on ethereum wallets and so we started we got a third person in lars who's the third co-founder actually lars and i first started working on it and we involved evgany afterwards and effectively these two guys of gany and lars are incredible data engineers. So they were able to architect the solution where we could grow out this wallet labeling at scale, very fast. And I'm a very results-oriented person. So I started tracking how many addresses are we tagging, how much volume are we covering out of all the transactional volume,

you see in Ethereum. And we were growing that really fast. And I understood that maybe we have something to work with here. But when we started testing it out as just a wallet labeling API and like a service, people didn't really know how to use it. So they were like, okay, this is cool, but how can I get something out of this? I would need to integrate this with my system and so on and so forth. So what we said was let's just combine this with the actual on-chain data. So let's just make one holistic analytics platform where not only do you get the vanilla on-chain data that you can pull out from a blockchain, which we would use Ethereum ETL for, again, is open-serve project, but you enrich that data with entity information and behavioral tags. And we quickly saw

just from using the product ourselves, and I still am a power user of the product. And I just saw that, you know, it just makes it so much easier to understand what happens on the blockchain when you have that the annotation and the attribution on top of the on-chain data. So yeah, that's how we ended at building nonsense. It's a really powerful tool because on one hand, one of the beautiful things about public crypto networks is that you have real-time data and transparency in a way that you've never had for any other asset class. And you have so much data. But then the challenge is parsing through that data and kind of separating the signal from the noise and making it actionable. So I think what you're doing and what you're doing and what you've done.

built is really impressive. Thank you. I mean, our mission is literally to surface the signal. So I think you're absolutely right. There's an abundance of data in the crypto space. Yeah. But the signal is pretty hard to surface. And if you think about something like Google as a search engine, obviously all the data for the websites is out there. But what they do as a service, as a search engine is to curate that data and to surface the stuff that you care about. And I think it's quite analogous actually to the blockchain industry in that sense. Yeah. Going back to wallet labeling, how do you go about getting started on that process of labeling wallets and how do you automate it so that over time that's not the bulk of your focus and you can focus on

kind of extracting insights from that process? As you alluded to most of the tagging is automated it because it wouldn't scale to do this fully manually. We tag up maybe 100 to 200,000 new addresses every single day, and that's algorithmic. But there is also a very important human component here because there's so much stuff happening in crypto. And if you create heuristics, algorithms, or even machine learning models, you do need to have a human in the loop. In fact, you need many humans in the loop.

And so what we do is really, we try to combine man and machine. we have one team called the attribution team, and at a high level, their job is to label up as many addresses as they can and to label so that we cover the most transactional volume. But this team has three legs. One is to build out the infrastructure and the actual heuristics that tag up wallets at scale automatically. And then on the other hand, you have coordination of people, to some extent,

who are trained to do tagging, but also there's a prioritization going on. Because let's say you have 20 humans working on tagging. You want to make sure they're looking at the right things. And so the third component is ready to build tooling for those moderators or people looking at addresses and to build good tools. And we actually use our own tool a lot to infer what certain addresses are and to understand the better.

But we also, of course, have a lot of, internal tools that are not customer facing that we use to track down. If we see like a press release of a funding round, we can often figure out the lead investor is this. And so the largest token transfer is this. And so we can sort of put two and two together. There's no silver bullet where it's like we have this individual algorithm that does this magical thing. It's more like we try to build out a system and we have people and processes such that we can do this at scale, but with very, very high precision. Could you talk a little bit about you have the on-chain data and then you have this labeling of wallets that's superimposed on the data? So what are some of the metrics and insights

that Nansen is uniquely positioned to provide? And then of those, I'd also love to know what from your perspective, your user's perspective, are some of the highest signal metrics that you and your clients use to regularly make trades and investments? Yeah, it's funny because I often get this question about what are the most useful metrics. And to me, actually, Nansen is more useful looking often at individual transactions. It's kind of like going from the very top level all the way down to an individual transaction that took place and that can inform you. that could be right now nfti collections are very hot and we have a lot of users who use nonsense

to invest in nfti collections and so one dashboard we have is nftt paradise which shows the top nfti collections in trading volume i can use that as an example but of course we have similar dashboards for just tokens erc 20 tokens and so on but the first step is to discover that hey there's something going on there there's a new nfti collection it was launched 20 minutes ago, there's a lot of ether being sent to the minting contract or where they're selling the NFTs in the primary sale. And then you can drill down and you can see, hey, who is actually contributing the ether? And so that's one step deeper. And you can see, oh, there's an NFT influencer who is minting a lot of these. That gives me strong conviction that this NFT collection might

become popular because they have a big audience. From there, you can choose to jump into the opportunity or not, but you can do this kind of due diligence one step deeper after you've discovered the opportunity. And then you can set up alerts on that wallet. If you wanted to track when this NFT inference is like starting to sell these NFTs, you'd probably want to know about it. So you can get telegram alerts or Discord or Slack alerts that you can set up in like five seconds from there. That's how the journey very often looks like when people use our product. of course there are also high-level metrics so we track things like how much ether has been deposited into ethereum two the staking contract or the deposit contract which lets you

stake on ethereum two that's like a very macro metric but we track stuff like that and we also break down who are the main contributors of eth into that which is uniquely something we can do because we have the best coverage on ethereum when it comes to wallet labeling i think there are more sort of well-known metrics, like how many tokens are sitting on exchanges. So in order to answer what percent of chain link tokens or what percent of USDC is sitting on exchanges, you need to have extremely good attribution of the wallets. Because if you don't have high coverage, you will underreport the numbers and you can miss out on important movements. And so I don't And I don't want to like dunk on any other products out there.

But I've seen examples where the incorrect conclusions are drawn because they don't have good enough coverage. And so let's say that you suddenly see, hey, there's a lot of chain link tokens like they're disappearing from exchanges or mattock tokens or whatever it is. But in reality, they're just being moved into Coinbase. And they don't have the good tagging on Coinbase. And so if you need to have extremely high recall or high coverage on labeling exchange wallets, and so that's one thing that we focus on, so you can look at these big picture metrics as well. Macro indicators, yeah. Exactly.

The thing is you can go from the macro view all the way down to individual transactions. Yeah. And, you know, I have a lot of anecdotes of people who have discovered stuff in Nansen. One of the first crypto funds that we had on the platform, I think the first, week they used it, they discovered that another fund was accumulating the same token that they wanted to accumulate. They concluded that they had to accelerate their own accumulation of it. And they said that they saved something like $250,000 in execution costs like estimated. And they paid $150 a month for our product. Of course, that doesn't happen with everyone.

We can't run and say that's going to be the case for every customer that used our product. but there's a lot of interesting anecdotes on how people have gotten value about it. What of my questions was going to be, you have other data and analytics firms like a chain analysis, like a glass node, like also going through the process of labeling wallets. And one of my big questions was going to be like, what are some of the key differentiators of Nansen? And I think you really answered it in that question.

inadvertently because, you know, I think a lot of them focus on the more macro indicators and try and draw out insights from that. That's right. But Nansen places a really massive focus on the individual transaction level and providing as much color around individual transactions as you can. That's exactly right. To avoid kind of misleading onlookers and participants to the wrong conclusions. That's exactly right. Yeah, I think that's a great way to summarize it. And I think there are two reasons for it. First of all, individual transactions do matter in many cases. And second of all, individual transactions have very strong narratives. And narratives are very important in crypto. So if you see, hey, three arrows capital just bought this token and that the word gets out on

Twitter, you would want to know about that as early as possible. And the blockchain is literally the first-hand source of information for that transaction. So that's why I think individual transactions are actually really important. And maybe as such, nonsense is kind of like half blockchain analytics or on-chain analytics and half blockchain explorer as well. So we kind of sit maybe in between those two. Another question I wanted to ask you, and it's a little different, I think, for each analytics company in this space, but who is your target? target customer segment or end user or who has gravitated to Nansen and maybe what's the breakdown? You mentioned funds, but what do some of the other customers look like?

As I mentioned before, our mission is to surface the signal. And if you look at that in a bigger scheme of things, we really want crypto and decentralized finance to basically take over for traditional finance. We think this is really the future of finance. But that doesn't happen by itself. You need to push that vision forward and you have to make it a reality. It won't just happen by itself. And we think one really important way that it happens is that if the pioneers in the space, if the people are on the front line of like defy and FTs and so on, if those pioneers emerge as winners, we want to see them succeed because they will reinvest in the space and they will also attract other people to join the crypto space. And so for us, it's important that

we can help them surface a signal because that's one very important way that they will emerge as winners. And so we focus a lot on the crypto natives. That's obviously a big difference between us and something like chain analysis, which has similar tech, similar data to some extent, but they sell to a completely different customer segment. A lot of it is actually in the public sector in which you probably know. So our customers are crypto investors. They are typically more sophisticated. sophisticated in the average. I don't say that just to flatter our customers, but it's just the way it works. We don't have a free product. It's all paid. We have a subscription-based business model. We have a lot of individual investors, Wales. We also have most of the crypto funds out there from the small ones to

the absolute biggest ones. The market makers that are active in the space, use our product, OTC desks. We have token listing teams from the biggest. exchanges who use our product because they want to understand the trends, what's happening on the blockchain, which new tokens are getting traction and so on and so forth. And yeah, so it's pretty much any pioneer, I would say, in the crypto space. And it's typically crypto-natives. But we are building out an institutional team as well. And so that's also going to be crypto-native at first, focused on institutional funds in the

space, and then we will expand towards the sort of more crypto-curious funds that maybe come from traditional finance. And then as more people enter this industry, we will hopefully be able to sort of shepherd them and provide them with high-quality information and analytics as they make that journey. That makes sense. Yeah, it's been great to see more institutions and just mainstream non-crypto-native firms take an interest. My prior to entering the crypto space, I was an equity research analyst.

And it's been great to see more equity research, sell side teams, take an interest in defy. I haven't seen a sell side report on NFTs yet, but I'm sure it's coming soon. It's good to know that teams like yours will kind of be there. It's a help guide them to the right information. Because at least initially, I feel like they've been gravitating to. be immediately available kind of free resources, but I think as they get more comfortable with the space, they'll be like, this is definitely something worth paying for that kind of gives us

an edge relative to other teams. I would also say that it's not a coincidence that we have focused on the crypto natives first. I do think that that's actually a unique advantage we have right now because our customers are literally on the front line. They're the pioneers. They have really high expectations on our ability to tag up addresses, to support new blockchains, to add bespoke dashboards for certain defy protocols. And I think that's going to help us as we expand to the less crypto-native segments, because we will have brought so much data and input from the pioneers. I think that's something that we really want to nurture. So I think we will always be a crypto-native first product, but we do also want to make sure that it is a product that other

customer segments can get value out of as well. And I think like the two will hopefully, eventually, the lines will blur between the two. So I'm looking forward to that point in our future. So right now, Nansen provides data for Ethereum, finance smart chain and Polygon. Could you compare and contrast the ease of supporting these respective layer one and layer two networks, both from like a technical perspective as well as from a parsing signal from noise perspective. And then, yeah, like is it vastly different or is it pretty similar across at least these three? So we have a chain evaluator framework, which is how we decide which chains we want to onboard. and the sort of short summary of that is cost benefit as most other things.

But on the cost side, for example, look at is this chain Ethereum virtual machine based or not? Because if it is, it facilitates the onboarding significantly because we can reuse a lot of our technology. So Polygon and BSC are both EVM based. And so we can actually reuse a lot of our tech. and we're actually right now building out even more scalable EVM support so we can faster onboard any EVM chain. But there are many chains out there that are EVM-based. So like Phantom, Ronan, Avalanche's C chain as well is EVM-based.

That's one point where we try to focus on the EVM chains first. The other reason for it is that those chains tend to have good bridges with Ethereum. and if they have good bridges, that means the users will very easily jump between them. And there's a higher chance that you're going to use Polygon than some other non-EVM-based chain because you can continue using Metamask and things like that. So it has a natural sort of overlap with users that are already our customers. But at a high level, we look at things like user feedback, you know, are people asking for this chain? And then we have strategic opportunity.

do we think that three, four, five years from now, this chain will have active usage. And then the cost component is also very important. Some things that come into play there are, I mean, I mentioned the EVM aspect before, which is mostly on the development cost. Do you have to write new code to parse out the data? And so if you look at something at Solana, that is completely different. So the cost is much higher for us to integrate it. But also the running costs can be quite substantial.

essentially different because the throughput is very different for these blockchains. So if you're supporting a blockchain that has like 15 transactions per second, that's very different from supporting one that has like 5,000 transactions per second in terms of how much data is generated. So I think Solana, for example, would generate something like 100 gigabytes of data every day, which is significantly more than something like Ethereum. that's something we have to take into consideration as well. And of course, automation when it comes to attribution and things like that becomes even more important when you have that amount of throughput.

So those are some of the ways we think about it. It's mostly cost benefit and we look at user input, strategic opportunity and then on the cost side, many of these different technical aspects. We also look at how well developed is the technical ecosystem around the chain. So are there third-party node providers that we can use so we don't have to run a node in-house? And then, of course, there's a financial component as well. So some chains will have grants where they can offer that, which, of course, makes the cost-benefit calculation. It can sometimes tip it in their favor. Yeah.

Exactly. Exactly. Because it is a chicken and the egg problem. Many of these chains, the reason they don't have that much activity is because you don't have good enough explorers and analytics tools. And so we can actually help these chains get more usage. And so I think the pitch there is pretty clear for any chain that hasn't like fully taken off yet. But it is not our intention to be supporting like a bunch of VC ghost chains either to use that term. So you have to balance out these different things.

What about the introduction of newer layer two networks on top of something like, Ethereum because that's the dominant smart contract platform right now. How does the rollout of like optimistic roll-ups or ZK roll-ups, how does that impact Nansen? Does it change anything? Is it going to make it more difficult to provide as rich and actionable data? How do you think about that? Yeah. Actually, arbitram and optimism are the two first layer two solutions. If we, there's a discussion on like is polygon layer two or not but I think arbitrament optimism everyone agrees are layer two solutions and we've pretty much treated them as like normal EVM chains and that's not a significant challenge in any way I think it's a little bit different if you look at something

like zK sync or starkware solution that's different and frankly I can't really comment on it because I haven't looked at those very closely yet arbitrament optimism we've been basically already have built support for internally and we have data parsed out. We don't provide it yet on the front end because there isn't a lot of activity yet. As soon as we see activity picking up, we can usher forward the rollout of L2 support for optimism and arbitram at least. And yet there's like side chains as well, like Ronin for Axi Infinity. Again, that's like EVM-based so we can sort of use the same technology to a large extent. You have to tweak it a bit. Sometimes the nodes have certain quirks that you have to deal with.

With BSC, we actually struggle quite a bit with the nodes because it's a really heavy chain. I think we worked a lot with our node providers to actually help them improve their offering because we needed higher performance. We needed to query the data super fast and we needed to up the rate limits. And Evgeny, one of my co-founders, he's built integrations for like probably 20 plus blockchains. not just EVM chains, but he's done this in the form of open source projects for many years now. So he obviously has incredible deep insight into this. And he can figure out how we need to tweak the nodes and also give that input to the third-party providers.

So, yeah, I mean, there's a lot of challenges around this stuff on its own. Attribution and labeling is one very important part of our business, but it's also not trivial just to like pull out this data in real time and make it easy. consumable and available to our customers. Yeah, that makes sense. Another thing I wanted to ask you was, so it's nansen.a.I, and you obviously have a background in AI. How has that played a role to date in your strategy and then longer term,

and more broadly, how do you kind of see blockchain data and AI merging in a way that's beneficial? Anyone who has worked with AI, machine learning, data science, they know that the most important thing you have to do is build a strong foundation. So you need to invest in your data infrastructure. You need to have excellent data engineers. You need to make sure that you have really clean data, you know, data pipelines with good QA processes and so on and so forth.

So to date, I would say we have not even scratched the surface, of what you can do with both machine learning and AI more broadly on blockchain data. We have done, of course, proofs of concepts. We have some cool stuff that's going to be maybe we'll share something end of next week. But obviously pricing models for NFTs is like an obvious thing to do with machine learning. It's totally analogous to what Open Door does with pricing real estate, for example. That's one example. We already have some models in-house that.

actually perform surprisingly well and we're going to refine them. We also, of course, have done proofs of concepts on using machine learning for the tagging itself, which is kind of tricky because on the one end is very promising. We have really high accuracy and precision if we try to, for example, build models that can recognize exchange deposit wallets. So like where you send your funds, if you send USDC or Ether to Binance, you'd have your own deposit wallet. So training models to recognize the transactional patterns. That's very promising. We have high high accuracy and so on with those models. But you also need to make sure that there's a probabilistic component involved when you do machine running. We haven't dedicated that

much time yet to making use of these machine learning models and get the human in the loop and the moderators to actually sort of sign off on some of the predictions or monitor the quality and the performance. So I think most of the exciting stuff with machine learning is definitely ahead of us. And when I speak to machine learning engineers and data scientists in interviews, people who want to work with us, I say this. And it's really true that pretty much all the interesting challenges are still ahead of us. And this is a really great time to join our company because we have built a really solid foundation. Two of my co-founders are data engineers. I mean, not that many companies have that privilege, and I'm a data scientist myself.

We are data people from the management level and down. I would say there's a lot of potential, and I'm happy to talk about use cases for it as well, because the pricing part is like an obvious thing. And you can do the same thing with, obviously with tokens too, although you have to use a different approach. Evgeny and I in 2019 built a proof of concept like token recommender. So a recommender system, not like financial advice recommendations, but the same principles that Amazon would use to recommend items to you based on your purchase history, you can do the same thing with on chain data and just say, if you bought zero X and Khyber tokens or whatever, then you're likely to also be interested in this other token. It's the exact same principle.

And to do that, you would use collaborative filtering or other recommender systems to make those recommendations. And so I don't think that the way you get AI into blockchain is to like literally put the AI on the blockchain, which I've kind of seen some boil the ocean, ICO white paper things kind of pitch. but using AI methods on on-chain data has a lot of potential. Right. And there's tons of applications that over the next 12 to 24 months, I think we're going to see many examples of this. Yeah, I think there's like a symbiotic relationship between the two.

So I'm definitely excited to see what you come up with there. And the pricing model for NFTs is interesting, especially because I feel like price discovery in the NFT ecosystem is non-existent. Yeah, the cool thing, if you have a good pricing model, one thing is that you can give the price recommendations to your users. It has a sort of direct utility, but also has indirect utility.

So you can suddenly start estimating market caps in segments of a collection. Like, what's the total market cap of this pool of penguins or whatever? And you can also say, here's the value of your portfolio. instead of just saying, here's the floor price on all of them or your last purchase price. So it's actually also useful as an infrastructure component if you have really good pricing models for NFTs. And so I'm pretty excited about that. I want to dig into NFTs because I know that you've personally been very excited about and Nansen has rolled out multiple dashboards on NFTs.

So to just kick off the discussion, what are some of the things that you highlight in your NFT dashboards and how are you or your customers using the dashboards to make decisions about what to participate in and what to avoid? It comes back to this framework of discovery, due diligence, and then defense of getting alerts when certain events happen. And so the way you discover NFTs, first of all, is typically through minting activity. So in the primary sales, there's three different places a sale might show up in Nansen. One of them is like NFT Paradise itself, which just ranks NFT collections. And you'll see, hey, there's a lot of transactional volume going on in this collection. And I see the smart contract was deployed like 22 hours ago.

and so boom, that's new. I want to look at that. That's one way. The other way is in hot contracts, which is a more generic dashboard that just shows capital inflow to smart contracts. And it also highlights the smart money deposits. So we have this aggregate segment that we call smart money.

This is like a few thousand addresses that have had good trading performance or investment performance in the past, or they belong to known like hedge funds, VCs and so on. And so we can see, hey, this smart contract has like $10 million in flow and three different smart money addresses put funds into it. That can immediately draw your attention to it. And sometimes NFT collections will pop up in that dashboard too.

And the third place is in our gas tracker dashboard. When there's a lot of minting activity that tends to spike the gas price on Ethereum, because people go crazy trying to make sure they get into the round. you can see the spike, but you can also see what's the top address that's contributing to the spike. And often that's the NFT smart contract itself. So those are like three ways you discover the NFT collection. And I've, I've done that myself. That's how I discovered hash masks, board apes during the mint. And sadly, I sold all my board apes too early. They would have been worth, I think, of like $5 million at this point. Yeah.

It's held out to them. But yeah, so that's how you discover them. And then on the due diligence, intelligence part, people typically, as I said before, try to figure out who else is putting funds into it. You also try to monitor how many unique addresses hold this collection. And of course, the higher, the better, because it means it's more distributed. And I think this is actually even a better metric than for normal ERC 20 tokens, because typically it corresponds better, like more one-to-one with individual people. With tokens, like ERC20 tokens, you get a bit more noise because you can have exchange deposit addresses that hold funds. So it's a little bit more noisy.

But with NFT collections, typically there's a stronger correspondence between persons and addresses. And so if you see a collection like board apes, when they flipped 5,000 addresses, that's like 50% of all the board ape tokens that exist. and so it's really evenly distributed across the holders. Cryptopunks is still, I think, maybe 3,000 holders. So it's actually quite a lot lower, even if it's been around since 2017. And yeah, of course, the pudgy penguins, which I've been a fan of publicly. I only own three, so I'm not shilling the penguins. I just think it's funny.

It's a fun collection with, like, lots of positive and friendly people in the community. And they're just super cute. Yeah, and also very cute. That's the fundamental value in those NFPs of course. So they're like close to hitting that tipping point of like 50% where there's 8,888 tokens. And like if you hit 4,44 addresses, that's like 50% distribution, right. Based on what you've seen is a smart money that is very active in Defi. Is there like an overlap between the smart money?

in Defi and the smart money in NFTs, or has it been an entirely new group of smart participants? That's an excellent question. We actually wrote the research piece on this. Oh, amazing. Because I think it's something that many people wonder about. Anecdotally, I think the groups are quite distinct. If I think about people I speak with and so on, like it's definitely true that you have people who do not care about anything other than NFTs and actually I would say this is more and more I like to say that DFI brings the capital and NFTs bring the people and I think a lot of the people who are interested in finance and they engage with DFI but if you go out and talk to like a random person on the street

they don't care about finance but they do care about art entertainment music gaming all these different things and that's why NFTs appeal to them. And so you have people who come from they've been trading like counter strike items or like World of Warcraft item in the auction house and they're like, wow, NFTs are really cool. It just clicks. Yeah. They get it, right? Like you don't have to convince them. They just can't get drawn towards it. And then you have other people who are just, they like the aesthetics, they like to support artists. So I think on the first part, anecdotally, there's definitely there's some degree of distinction. There are people who are only interested in NFTs.

I think it's also true to the flip side. There are some DeFi folks who are a bit more like, I don't know if rational, that's the right word, but more like they care about things like cash flow yields, returns, and they have no interest in the sort of aesthetics. Like more fundamental driven, yeah. Exactly, like the community aspect. But I think more and more I see people in that camp also getting it

with NFTs. And I also feel there's almost a bit of an analogy to the way traditional finance people think about crypto. I think sometimes some of the defy people, that's how they think about NFTs. It's sort of like in the beginning, they think it's just silly. They don't get it. And then there's something that clicks and they're like, hey, there's something bigger here than just silly JPEGs, right?

There's something more to it. And then they get it. But yeah, so that's the anecdotal side. on the data side, there is definitely overlap between highly active dext traders, for example, which we looked at and we tag up and then highly active NFT collectors. And I think there are some practical reasons for it. As you use, say, metamask and stuff like that, it's natural that you also know how to use uniswap. So there is some overlap in activity there, maybe a bit more than we expect it.

I don't have the numbers in front of me, but I think if you Google for something like Nomsen, Defi, NFT overlap or something, I think you'll find the research piece. Young and Paul, two researchers, and our team looked into this, and they literally just looked at addresses that are active in NFT trading and at different activity levels. Some are highly active. Some are just a bit active. And the same thing on like the decks trading side. How much do you trade? Are you a heavy dext trader or an elite dex trader, as we call top ones? Or are you just sort of like a more casual? dextrater. We have a matrix where you can see the overlap between those two different

behavioral segments. And something that I think I and a lot of people are curious about is you have a lot of obviously individual people participating in NFTs, individual whales. But what about funds? Have funds, have many funds been really involved and like aped into different NFT collections? Yeah, more and more. We have funds. We have funds. We have come to us asking us if we can help them with basically procurement of NFTs. So I obviously can't reveal names, but these are big funds that are seriously looking at NFTs. They need advice on basically the whole journey. So NFTs in this sense are a bit more like acquiring real estate than they are like acquiring crypto because you need to figure out what collection should you be looking at. And then

within that collection, what are the items that we believe will retain value or be rare exactly? Like, what does the market think about white hair punks, for example, or trippy fur board apes? You need to have some domain knowledge about the different collections. And then, so let's say you figure out, hey, you want to buy these pieces. Well, most likely those pieces are not for sale. only a small fraction of NFTs will be listed on, say, OpenC or other platforms. So is there a way we could reach out to them and actually get touch with the owner? And so that's where NAMSEN becomes very important again, because we have a really good

overview of the different owners. In that sense, it actually becomes quite useful for OTC purposes. So certainly there's interest, both on the fund side, but also on whales that need help with OTC, purchases of NFTs that they can necessarily like find or reach out to the owner of a specific piece. That's a big unsold problem. We have ideas on what we want to do, but I definitely foresee more projects trying to address that problem. You also see other examples of very sophisticated executions of purchases where you can see I think Ivan from Dragonfly was was helping someone and make a large, basically sweep the floor, as you call it.

You sort of batch purchase NFTs so you can't be front run. You try to buy them all in one go. For now, a lot of bespoke tools have to be built to make those transactions. But clearly, those are getting at productized, I think. And so I think funds, if they want to go big into NFTs, they will be making use of those kinds of tools as well. And of course, there's a visa purchase of crypto-pominy. which many of the listeners will know about.

This is definitely being looked at from big funds, and it's not only an individual market anymore. It definitely also has interest from large players and funds out there. I think that's very validating. And I think as that happens, we'll also see the blurring of lines between NFTs and DFI, because those tools will be built. built out to cater to the more institutional fund crowd that needs tools on that to participate

and do what they really need to do for it to have an impact based on the amount of capital they have. That's right. I think there's also an interesting discussion on the regulatory aspects of NFTs. And so I'm not a lawyer, so I don't want to make any strong statements here. But I think in many jurisdictions, NFTs fall outside of both securities regulations. and also outside of, for example, here in Singapore, we have digital payment tokens regulations. At least as far as I understand, having spoken to a few different law firms, NFTs do generally not fall under any of those. And so that makes it, I think also more attractive for startups to build solutions for it, because you have a bit more room to create interesting applications without worrying too much about the regulatory aspects.

And also investors who want to engage. with it. So I think there's a kind of like, we often say that crypto has a bit of regulatory arbitrage versus traditional finance. I think NFTs have a bit of regulatory arbitrage versus the rest of the bill. Yeah. So I think that's another, maybe another reason why some institutions are getting more interested in NFTs as well. So we only have a few minutes here. So I want, I want to get your thoughts on just the broader data space. How do you think about crypto data and analytics longer term, do you think there's going to be many providers that are really good at their specific niche, or do you foresee continuation of consolidation and M&A, either by different data

companies that kind of merge together or by crypto exchange and brokerage type businesses that end up leveraging data and analytics as a value ad for their customers? So I'm 100% sure there's going to be a lot of M&A. M&A activity. I don't think chain analysis has raised hundreds of millions of dollars to run Super Bowl ads. So definitely on that one within a year, I would foresee if at least one big acquisition, right? Like maybe more. Then I also think it's true that there are crypto, because it has so much open data and an abundance

of data, I do think that there's always going to be this long tail of interesting analytics websites. and people making cool stuff. So I think one good example was with EIP-1559, which was a part of the London upgrade with Ethereum. There were a lot of websites that popped up around sort of the burning of ether. And we actually sponsored one of them, Watchtheburn.com, which is a cool site. It's a really popular one, yeah.

Yeah. And the cool thing is individual developers can actually build out many of these niche websites that are focused on something very specific. So I think we're going to continue to see more of that. It's one of the reasons I got attracted to crypto in the first place. As a data person that you can actually make all these cool applications or dashboards and stuff like that. But yeah, so I think you're going to have this continuous inflow of new cool projects. Speaking openly, on our side, we also want to engage with these builders to do this stuff.

So we love sponsoring these different websites. we don't see them as like competitive at all. In some cases, I could also foresee that we would make offers and acquisitions to some of these smaller teams. We don't have the financial muscles that Chen Alsace has. So like we can't make the biggest acquisition offers yet. But I think we can certainly present a compelling offer to many of these builders who want to be in a larger family of data enthusiasts where you can actually focus on. on the stuff that you are passionate about, which is data analytics,

and you don't have to worry about going out, like, I don't know, raising funds or, you know, selling ads on your website or like whatever it is that you are doing now. And then, of course, our data as well can 100x the value of what they are building. If you have a website and it's like, hey, actually this dashboard or this chart

would be extremely powerful if it had labels on them or like entity information, and so we could make it even better. So I think the short answer is like, I think we're going to see a lot of MNA activity at the whole scale from the small to the big ones. But I think there's going to be a consistent inflow of great data analytics people building cool websites in the space. I definitely agree. So Alex, we're at the top of the hour. It's been great to have you.

Just as a last question, what's the best way for listeners to kind of follow your work? Where can they find you? where can they find out more information about Nansen? You could give us a little bit on that. The best thing is to go to nansen.a.i. That's our website. You can do the trial for, I think it's $9 for seven days, not too expensive. You can also follow me on Twitter, A. Svanevique.

So that's A and then my last name on Twitter. Just look for a penguin. And yeah, that's the best way to find me. Awesome. I will also plug your YouTube channel. I've been following along your office hours for the last few weeks, which have been great. So lots of good alpha in there for anyone that's interested. Yeah.

Thank you so much, Alex. Many people like that channel, so appreciate you for watching. Yeah, of course. Thanks very much for having me. Thanks for listening to another episode of On the Brink with Castle Island. To find out more about Castle Island, visit Castle Island.V.C. To listen to all of our podcast episodes, please go to On the Brink-Podcast.com or just click on the tab in our website.

Thanks for listening.